Crafting Interpretable Embeddings for Language Neuroscience by Asking LLMs Questions

Large language models (LLMs) have rapidly improved text embeddings for a growing array of natural-language processing tasks.